Artistic Data Visualization: Beyond Visual Analytics
Total Page：16
File Type：pdf, Size：1020Kb
Artistic Data Visualization: Beyond Visual Analytics Fernanda B. Viégas and Martin Wattenberg Visual Communication Lab, IBM Research, 1 Rogers St, Cambridge, MA 02142, USA {viegasf, mwatten}@us.ibm.com 109 Homes for Sale, 117 Homes for Sale, 124 Homes for Sale, Seattle/Tacoma Chicagoland The 5 Boroughs 121 Homes for Sale, 114 Homes for Sale, 112 Homes for Sale, LA/Orange County Dallas/Ft. Worth Metroplex Miami-Dade County Fig. 1 Homes for Sale, Digital C-prints by Jason Salavon. Abstract. Information visualization is traditionally viewed as a tool for data exploration and hypothesis formation. Because of its roots in scientific reasoning, visualization has traditionally been viewed as an analytical tool for sensemaking. In recent years, however, both the mainstreaming of computer graphics and the democratization of data sources on the Internet have had important repercussions in the field of information visualization. With the ability to create visual representations of data on home computers, artists and designers have taken matters into their own hands and expanded the conceptual horizon of infovis as artistic practice. This paper presents a brief survey of projects in the field of artistic information visualization and a preliminary examination of how artists appropriate and repurpose “scientific” techniques to create pieces that actively guide analytical reasoning and encourage a contextualized reading of their subject matter. Keywords: Visualization, Art. 1 Introduction Data visualization is usually viewed as a tool to support analytic reasoning. This reflects its roots in science—ranging the economic charts of William Playfair to Richard Feynman’s diagrams for performing calculations in quantum field theory. Researchers in the scientific visualization community have naturally focused on supporting scientific analytic tasks. In recent years there has been a renewed interest in infovis’ role as an intelligence tool for military and national security applications. (This too has historical roots: several visualizations of war have become touchstones for information designers: e.g., the 1869 chart by Minard of Napoleon's march to Moscow and Florence Nightingale’s diagram of patient mortality in military field hospitals.) Visual analytics, the combination of infovis and mathematical deduction to extract patterns in massive, dynamically changing information spaces has, of late, become one of the main themes of the visualization academic community. As with research that supports scientific visualization, the emphasis has been on visualization as a tool for dispassionate analysis. This relentless focus on visualization as a neutral tool may appear to be a forced move. At first, bias in a visualization might seem like a technical problem, much like chromatic aberration in a telescope. Yet a recent, separate stream of thought in visualization calls this assumption into question. In the past decade a second very different type of information visualization has flourished. Art based on data has been featured at institutions such as the Whitney Museum of American Art [6] and the San Francisco Museum of Modern Art. [3]. Often such artwork is directly based on techniques first explored by the academic community. At the Austrian Ars Electronica art festival in 2004, a visitor could see interactive treemaps in one exhibit [20], play with an installation by Brad Paley based on social network analysis [12], and hear Josh On talk about his use of graph drawing a discussion panel. While the basic techniques used by these artists would be familiar to those in the scientific visualization community, their motivations and creations are new and different. In this paper we explore the implications of the surge of artistic interest in visualization. In section 2, we define our terms and describe some useful context. Section 3, the core of the paper, consists of a set of analyses of several prominent examples of artistic visualizations. Finally, in section 4 we make some suggestions about what the academic visualization community might learn from artists on visualization. 2. Definitions and Background Defining what constitutes “artistic” visualization is hard, if only because defining art itself is hard. To sidestep that philosophical question, our working definition in this paper is that artistic visualizations are visualizations of data done by artists with the intent of making art. This definition may seem like a tautology, but in fact it specifies a coherent and interesting class of work. First, the artworks must be based on actual data, rather than the metaphors or surface appearance of visualization. Many artists have used diagrammatic imagery as a base for their projects. One example is Simon Patterson, whose “Great Bear” relabels the famous map of the London Underground [12], and whose “J.P.233 in C.S.O. Blue” [14] appropriates imagery from a map of airline routes. While Patterson’s work has many merits, it cannot be said to be data visualization since there is no underlying mapping between data and image. A second point is that our definition avoids the issue of beauty: we do not contend that beautiful scientific visualizations are automatically artistic, or that visualization art must be pretty. Thus a microscope photograph taken as part of a scientific experiment would not qualify under our definition: no matter how beautiful the colors, the photograph would lack artistic intent. As the examples below show, focusing on intent rather than surface aesthetics provides a coherent category of work with important distinguishing characteristics from scientific visualizations. A natural question is why this new type of visualization (or this new style of art) has emerged. Any answer to this question is speculative, but two particular factors are relevant. First is the emergence of software tools that are appropriate for artistic production of data visualizations. Today one does not require a supercomputer or fluency in C++ to create visualizations. Instead, it is possible to create sophisticated visualization software using cheap hardware and friendly development environments such as Flash (www.adobe.com/) or Processing (www.processing.org). A second factor is that data has become part of the cultural discourse on several levels. Thanks to the internet, complex data sets such as SEC filings are available with a few clicks. Indeed, the internet itself can be viewed as a massive database. Moreover, government and corporate collections of data now play a critical role in the lives of citizens of many nations. As a result, it is natural that artists want to grapple with the issues raised by the controlling power of data. 3. Artistic Visualization Projects To understand the issues raised by artistic visualization, it is helpful to have a set of concrete examples in mind. In this section we provide descriptions of projects that have been successful in the artistic community and that use sophisticated visualizations of data. This is, by no means, supposed to be an exhaustive survey of the area—such an undertaking is well beyond the scope of this paper. Instead, we have chosen to focus on a purposeful sample of projects that highlight some of the central qualities of artistic visualizations. 3.1 Jason Salavon and the Power of Colored Pixels If information visualization enables the viewer to see unexpected patterns in a body of data, Jason Salavon’s art pieces confront the viewer with inescapable, pervasive patterns in everyday life [15]. From innocuous mementos such as high-school year books (Fig. 2) to racy centerfolds of adult magazines (Fig. 3), Salavon blurs individual pieces to focus our attention on the collective aggregation of human experience. Take, for instance, Homes for Sale (Fig. 1), which shows a series of realtor photos of single-family homes for sale in different cities around the U.S. Each piece encompasses a collection of homes on the market in a given metro region in the median price range for that area. The images are constructed by taking the mean averaging color of every photo, pixel by pixel. The result is a blurred view of an area’s weather pattern and ghostly images of the houses for sale. Miami boasts the bluest sky whereas Dallas has the greenest grass. Seattle, on the other hand, seems awash in an assortment of gloomy grays. Salavon utilizes the same averaging technique to explore a wide variety of themes that permeate visual culture. The Class of 1988 and The Class of 1967 (Fig. 2), for instance, are part of a series of pieces investigating rites of passage—in this case, high school graduation—and the conventional visual mementos that get produced to celebrate such events. The Class of 1988 The Class of 1967 Fig 2. The Class of 1988 and The Class of 1967, by Jason Salavon The 1960s The 1970s The 1980s The 1990s Fig. 3. Every Playboy Centerfold, The Decades (normalized), by Jason Salavon In Every Playboy Centerfold, The Decades (Fig. 3), the colored images confirm the formulaic compositions of the adult industry and the change in taste over the years. Skin tones and hair color get lighter as time goes by. To be certain, the technique of averaging the color of pixels in a collection of images is not new. In fact, computer scientists have utilized the same mechanism in a variety of applications in the field of image processing [18]. Whereas the intent in most of these applications is to use pixel manipulations as input to face recognition algorithms, Salavon’s subverts this original goal. The artist exposes the technique as the output of his piece by letting the collection of individual images dissolve into a field of color that carries meaning in itself. Expanding on the theme of pure color, Salavon has also explored the concept of narrative through the arrangement of colored pixels in an image. For The Top Grossing Film of All Time, 1 x 1 (Fig. 4), the artist digitized the movie Titanic in its entirety and extracted individual frames.